WAX: A High Performance Spatial Auto-Correlation Application

Abstract

We describe the algorithms employed by WAX, a spatial autocorrelation application written in C and C++ which allows for both rapid grouping of multi-epoch apparitions as well as customizable statistical analysis of generated groups. The grouping algorithm, dubbed the swiss cheese algorithm, is designed to handle diverse input databases ranging from the 2MASS working point source database (an all sky database with relatively little coverage depth) to the 2MASS working calibration source database (a database with sparse but very deep coverage). WAX retrieves apparitions and stores groups directly from and to a DBMS, generating optimized C structures and ESQL/C code based on user defined retrieval and output columns. Furthermore, WAX allows generated groups to be spatially indexed via the HTM scheme and provides fast coverage queries for points and small circular areas on the sky. Finally, WAX operates on a declination based sky subdivision, allowing multiple instances to be run simultaneously and independently, further speeding the process of merging apparitions from very large databases. The Two Micron All Sky Survey will use WAX to create merged apparition catalogs from their working point and calibration source databases, linking generated groups to sources in the already publicly available all-sky catalogs. For a given 2MASS source, this will allow astronomers to examine the properties of many related (and as yet unpublished) 2MASS extractions, and further extends the scientific value of the 2MASS data sets.